Decentralized Estimation of Forest Fire Spread Using Mobile Sensors

This paper presents a study of the ability to build an observer for a complex system using a decentralized multi-agent system for the coordination of mobile sensors. The environment is modeled using a CA model representing forest fire spread. The initial distribution for the different species in the vegetation is generated using a Perlin algorithm. Implementation is realized on GPGPU. A coherence measure for the observation error is defined. The observation itself is realized with mobile sensors and a decentralized coordination of the trajectories. We analyze the balance between individual and collective behaviours of agents which is required to achieve the best performance with respect to the chosen coherence measure. The collective behaviour of the mobile sensors is based on pheromones usage, inspired from ants behaviour.